Many theories of democratization suggest that extending the right to vote will lead to increased government expenditure (e.g. Meltzer and Richard, 1981; Lizzeri and Persico, 2004; Acemoglu and Robinson, 2000). However, these models frequently assume that government can engage in transfer expenditure, which is often not true for local governments. This paper presents and tests a model in which government expenditure is limited to the provision of public goods. The model predicts that the poor and the rich desire lower public goods expenditure than the middle class: the rich because of the relatively high tax burden, and the poor because of a high marginal utility of consumption. Consequently extensions of the franchise to the poor can be associated with declines in government expenditure on public goods. This prediction is tested using a new dataset of local government financial accounts in England between 1867 and 1900, which captures government expenditure on key infrastructure projects that are not included in many studies of national democratic reform. The empirical analysis exploits plausibly exogenous variation in the extent of the franchise to identify the effects of extending voting rights to the poor. The results show strong support for the theoretical prediction: expenditure increased following relatively small extensions of the franchise, but fell following extensions of the franchise beyond around 50% of the adult male population.
It is perhaps too quick a jump from 19th century England to contemporary advanced economies. Still, it is an interesting hypothesis that the current thinning out of the middle class will decrease the political support for infrastructure investment.